# Bayesian multilevel step-and-turn models for evaluating player movement in American football

**Authors:** Quang Nguyen, Ronald Yurko  
**arXiv:** [2603.17866](https://arxiv.org/abs/2603.17866)  
**Date:** 2026-03-18  
**Summary page:** https://www.summarizepaper.com/en/paper/2603.17866v1/

## Key points

- Utilization of player tracking data in sports analytics has led to significant advancements in evaluating player performance
- Quang Nguyen and Ronald Yurko introduced a generative framework for assessing frame-by-frame player positioning by comparing it against hypothetical alternatives
- Research focused on analyzing movement patterns of individual ball carriers in the NFL
- Developed Bayesian multilevel models targeting step length and turn angle at the frame level
- Conducted posterior predictive simulations to generate hypothetical steps taken by ball carriers during gameplay
- Framework applied to tracking data from the initial nine weeks of the 2022 NFL season to derive novel player performance metrics
- Showcases a sophisticated methodology for evaluating player movement in American football using advanced analytics techniques

## Summary

The utilization of player tracking data in sports analytics has led to significant advancements in evaluating player performance. Quang Nguyen and Ronald Yurko introduced a groundbreaking generative framework for assessing frame-by-frame player positioning by comparing it against a distribution of hypothetical alternatives. Their research focused on analyzing the movement patterns of individual ball carriers in the National Football League (NFL). They developed Bayesian multilevel models targeting two key components of player movement at the frame level: step length and turn angle. By implementing these models, they conducted posterior predictive simulations to generate hypothetical steps taken by ball carriers during gameplay. This innovative approach allowed for a comprehensive comparison between observed player movements and simulated alternatives, utilizing common valuation measures in American football. Applying their framework to tracking data from the initial nine weeks of the 2022 NFL season, Nguyen and Yurko derived novel player performance metrics based on this hypothetical evaluation. Overall, their research showcases a sophisticated methodology for evaluating player movement in American football and highlights the potential for leveraging advanced analytics techniques to gain deeper insights into athlete performance and strategic decision-making within sports.